There's a lot of talk about this... I like to jokingly say: "It will never replace genuine stupidity." AI is a buzzword in many products, especially tech related. The HoverAir X1 I have says its flight modes are "artificial intelligence" based; yet it will essentially do the same thing each time. I wouldn't call that "AI;" it seems like programming to me. AI implies (to me) that the programming will analyze the situation and react accordingly.
But "Artificial Intelligence" sounds more impressive.
AI is a real concern in some industries: you can tell the computer to write a song that sounds like Bruce Springstein... and it can do it. Or, write a novel in the style of Stephen King. And therein lies the issue: AI can read and absorb, even learn a style - is the output it delivers of the same heart and soul of the original creater? It has people in the creative industries up in arms: AI can replace writers, actors, song writers, painters, and do photographic retouching that is amazing... we've all seen the images on-line that are flawless... and don't resemble the original subject.
Yes, as a professional photographer, it was my job to make people look as good as they could; but without changing their appearance to the point where it isn't a representation of the real subject. Retouching? Sure - those blemishes are transitory; the wrinkles are dependent on lighting (and we took great care in the camera room to light each subject to show them off at their best). When digital imaging entered the scene, we had to re-evaluate how far to go with the retouching. I used to joke with the real estate people and politicians: "How many years and how many pounds do you want me to take away?" It needed to be a tasteful balance. In the days of film, a good photographer could light the subject, but retouching on the negative was needed to remove those "transitory" things. With digital, that all changed. And now, with AI, it is possible to "deep fake" an image, creating a situation that never happened.
Honestly, I'm glad to be out of that industry, and I know how much to NOT trust the old saying: photos don't lie. Image makers today don't need to know lighting like we did in "the good ol' days."
But, I digress (as usual). Any regular reader here knows my passion for riding. And the fun I have capturing videos of the various rides. For the last 4 years, I have used 360 cameras to add a fun look to those videos... you get a 360º view of what's happening, and you use video editing to select the "flattened view" on typical video. It is how I am able to get a pan-shot of looking forward from the bike, then swinging the view around to show me riding the bike. You cannot do that with a typical action camera; and before 360 cameras, you would have needed a multi-person film crew to get that video shot.
The downside of 360 cameras: the video editing is very labor intensive - for a 10 minute video, I may have several hours of going through the video, selecting the view, editing an hour (or more) of raw footage down to 10 minutes, then moving it all to a more typical video editing software to make cuts, add transitions, add captions, then even more time selecting music (that isn't copyrighted) and/or a voice over.
I do it because I enjoy the ride... and it's nice to look back at it when I can't be riding. I post it to YouTube because it is easy access from that platform: I can post it on my blog, on Facebook, and on the various bike forums where I participate. I don't do it for the "likes" or views; my blog and anything I do on YouTube is not monetized. And when I get a new bike or a new camera, I tend to shoot a LOT of video. Sometimes, I enjoy the creative process of making a video... sometimes, I run out of steam part way through (and just post a couple still images). I don't always have the passion for creating a finished video that I have for getting out for a ride.
Enter the new Insta360 X4 that I got a week ago. In form and function, it is pretty similar to the X2 that I have been using for the past 3 years (that is ancient in digital video technology years). The X4 is capable of 8k video; that's the entire sphere, so the end resolution is much less than that. Is it good? Well, it gives people on the internet a lot to argue about: "The resolution isn't as good as my GoPro, DJI action cam... blah, blah." No one is making big screen productions with these little cameras, but a lot of people like to argue about the minutia... most people are looking at the finished product on a phone or a tablet, so the resolution of 2.7k vs 4k vs 5.7k vs 8k is just fodder for "my camera can beat up your camera."
I don't really care. I like to have the memory of the ride. I won't be watching it on an IMAX size screen. Neither will 99+% of the people arguing about it. I almost always go for the least resolution that will allow me to see reasonable detail... otherwise, you are filling up endless external hard-drives for some moderately silly videos. Same with the still photos I post here: they are reduced a LOT from the original file size... it is something I have done since first being able to post a photo on the internet. If someone says, "It looks mushy when I enlarge it or print it." Good - I don't post images (or videos) for other people to enlarge, print, or otherwise copy.
So, now... artificial intelligence. Today, while out for a ride, I shot some video with the new X4. What? You're not surprised? But, after spending almost 2 hours editing and downloading the video clips - and that is before putting those clips into video softward to create a "finished" video, I decided to give the AI that Insta360 is promoting. How does it stack up against an actual human being doing the editing?
Thanks for asking. The AI does a better job with a smooth flow of keyframes (the process where you determine what the view will be). It is essentially analyzing each "frame" of video and making the movement flow smoother than I would have the time or patience to duplicate. But, the AI thinks I am part of the "Selfie Generation" that wants video of themselves, so it tends to find a subject (in this case: me on the bike) and make it the main thrust of the video. I prefer to show off the road and the ride... and occasionally get myself in there. In fact, on the videos I shoot with the Go 3 (not to be confused with a GoPro), they are only looking where the camera is pointed, which is generally straight ahead - the reason for a helmet-mounted camera. Some moto-vloggers set up 3 cameras, where one or two are pointed at them, and one is pointed on the road ahead. I used 3 cameras last week for a ride: the X2 getting the 360º view, the Go 3 looking straight ahead, and my iPhone for photos of the bike and the camera set up on the bike. It was way more work than I would care to do again... not necessarily for the editing, but for the time it takes when getting ready to ride - I like to strap on a 360 camera and go.
I was also testing a new feature of the X4: gesture control - you can hold your palm up and it will start (or stop) shooting video. Make it a "peace sign" and it will take a photo. Yes, it works. Sorta. It doesn't recognize a black motorcycle glove as that palm...
Can't say that I am a fan of riding with only one glove on. So, I will continue to use my phone as a remote for the camera. And a couple still images, the road and the rider...
What about that AI? Thanks again for asking. Today, I spent a couple hours going through the video clips, putting in keyframes, then downloading. It took a couple hours, and I don't have a finished video, yet. The AI? As a test, I loaded 4 unedited video clips and let it rip. About 2 minutes later, it had this...
No comments:
Post a Comment